data-marketplace-residential-service

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow ingests outsider-authored free text via the user’s search input (typing into #search-input) and then issues Data Marketplace name-search queries using that raw text (window.api.searchComplex(q) → /api/complex-search), so an outsider can supply poison strings that the workflow consumes to select results.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly requires the agent to read the live product index at https://datamarket.bigvalue.ai/llms.txt at runtime (SKILL.md instructs this), which directly controls prompt decisions, and the bundled web template loads third‑party map SDKs that fetch and execute remote JS at runtime (e.g., https://dapi.kakao.com/v2/maps/sdk.js?appkey=${C.KAKAO_MAP_KEY}&autoload=false, https://oapi.map.naver.com/openapi/v3/maps.js?ncpKeyId=${C.NAVER_MAP_CLIENT_ID}, https://unpkg.com/leaflet@1.9.4/dist/leaflet.js), so these external URLs are required runtime dependencies that can control behavior or execute code.